Develop of A Torque Feedback Interface for Virtual Acupuncture

Training

08118118, Shuang Peng, School of Automation

1. About

Haptic devices provide a straightforward method for mapping simulated forces to the user. This

enables the user to actively interact with the virtual environment. This makes them useful in

2. Hardware

The acupuncture process mainly involves two motions, insertion of the needle and rotation in the

tissue. So the haptic device needs to render insert force and rotation torque in order to simulate the

acupuncture process on real-world patients. However, most existing commercial haptic devices are

equipped with 3 active translation DoF and 3 passive rotation DoF. Such as 3D systems’ Phantom[1],

Falcon[2], and Omega[3]. For those with additional active rotation feedback hardware, motors and

transmission systems are directly mounted on the end-effector of the device. As a result, additional

weight and inertia can impact apparently on user’s experience. To address this issue, some solutions

compensated the tip’s weight on the software level with device’s dynamic model[4]. This project

proposed a tendon-based flex transmission structure that allows the tip’s actuation motor to be

The design of the tip follows these guidelines.

Weight – Weight of the tip must be limited, without cause obvious changing on device's dynamic

property. Also, it should be easy and comfortable to operate by users.

Workspace – The added tip should not interfere device’s original workspace.

Transparency – Tip’s actuation system should be low friction, low inertia, and highly back-drivable,

which renders the feedback torque directly to users.

Control and sensing – The tip should be torque-controllable with an encoder measuring needle’s

real-time position and angular velocity.

A. Flex transmission system

Fig 2 CAD drawing of the moving tip

B. Rotation clutch

The diameter of a common acupuncture needle ranges from 1mm to 3mm. Smaller diameter means

users’ fingertip skin is easier to deform when twisting the needle. As a result, rotation friction is

easier to be perceived by users. To reduce the free rotation friction, a clutch mechanism is developed

to disconnect the needle when there’s no feedback torque, shown in Fig4. The clutch consists of 3

needle. Under retract state, the clutch is able to bypass torque up to 0.4 Nm.

C. Actuation motor

Actuation motor is the defining factor of haptic systems’ performance. Most existing applicants use

coreless DC motors with planetary gear reducer. However, geared reducer would introduce

unwanted friction to the system. The actuators used in this project are 1613 BLDC motors with a

KV value of 130, which provides continuous torque up to 0.08Nm under direct drive configuration.

Although this torque is lower compared to most geared coreless motors, they’re adequate for

generating rotation feedback with small diameter needles. Also, the output torque of the BLDC

motor has a good linear relationship with the phase current. This enables precise close-loop torque

control with current sensing embedded in the controller. Fig 6 shows the motor used in this project.

In acupuncture simulations, the needle’s position and velocity must be precisely measured in order

to calculate simulated feedback torque. Both tip’s rotation shaft and actuator motor are equipped

with magnetic encoders shown in Fig 7. Encoders are AMS5600 from austriamicrosystems AG with

a resolution up to 0.088°.

E. Controller

A customed controller board is designed as shown in Fig 8. The board contains 2 BLDC motor

drivers with phase current sampling, 2 IIC ports to read encoders, and a CAN bus to communicate

with the host computer. A dedicated control loop running at 1.5kHz sends measured needle position

and velocity to the simulation, then receives calculated feedback torque to actuate the clutch and

needle.

With the pose of tip, the insertion depth of virtual needle in tissue can be obtained, shown in Fig 9.

The target feedback torque can then be calculated by the following algorithm.

The measured result is shown in Fig 11. From the plot, it can be seen that the output torque is

basically useable. The noise on the signal can be improved via fine-tuning of the motor controller’s

paraments.

Fig 11 0.02Nm step output and sine output

4. Conclusion

This report presented the development of a lightweight, highly backdrivable torque feedback tip for

time, a compact controller is developed to control BLDC actuation motors. With a designed

evaluation, the system is able to deliver desired feedback torque calculated by the host computer.

The future work will concentrate on the feedback control of the clutch mechanism through a new

design with a pressure sensor. This will allow applications where controllable dumping modules are

References

dynamic identification and model feed-forward control of Novint Falcon haptic